First, Ben's suggestion to split grading vertically rather than horizontally - questions rather than students - is good and practical irrespective of anything else.

That being said, I suggest you consider the following:

1. Give up the fantasy, or perhaps better put: the conceit, that grading is consistent. It isn't, due to a zillion factors: Exams not of consistent difficulty; student population distribution of capabilities not in sync with grading curve shaping; sense of severity of errors differs for the same person for the first encounters and for latter encounters; moods change; unconscious biases (for example: against people with sloppy handwriting); and so on.
2. Try to schedule a grading policy discussion of all graders and the exam authors, before the next exam is given (next semester?). In that discussion, bring up some specific scenarios requiring finer judgement rather than simply "how many points for which question".
3. Consider arguing for rougher-granularity grading in the course. I'm a fan of Pass/Fail (or Pass/Fail/Excellent), and dislike number scales, especially 0..100 or fractional grades. I can be confident about looking at someone and saying "Yeah, you understand what we've taught here, you pass." or "No, you don't get it - you fail." And if someone is borderline, then it's a matter of policy whether to pass or fail them (I would tend towards fail personally). But I really cannot justify why someone is a 63 while another person is a 64.7. I feel I'm only helping some arbitrary industrial mass-manipulation mechanism by assigning these kinds of numbers to people.

PS - These three suggestions are mostly orthogonal.

First, @Ben's suggestion to split grading vertically rather than horizontally - questions rather than students - is good and practical irrespective of anything else.

That being said, I suggest you consider the following:

1. Give up the fantasy, or perhaps better put: the conceit, that grading is consistent. It isn't, due to a zillion factors: Exams not of consistent difficulty; student population distribution of capabilities not in sync with grading curve shaping; sense of severity of errors differs for the same person for the first encounters and for latter encounters; moods change; unconscious biases (for example: against people with sloppy handwriting); and so on.
2. Try to schedule a grading policy discussion of all graders and the exam authors, before the next exam is given (next semester?). In that discussion, bring up some specific scenarios requiring finer judgement rather than simply "how many points for which question".
3. Consider arguing for rougher-granularity grading in the course. I'm a fan of Pass/Fail (or Pass/Fail/Excellent), and dislike number scales, especially 0..100 or fractional grades. I can be confident about looking at someone and saying "Yeah, you understand what we've taught here, you pass." or "No, you don't get it - you fail." And if someone is borderline, then it's a matter of policy whether to pass or fail them (I would tend towards fail personally). But I really cannot justify why someone is a 63 while another person is a 64.7. I feel I'm only helping some arbitrary industrial mass-manipulation mechanism by assigning these kinds of numbers to people.

PS - These three suggestions are mostly orthogonal.

First, Ben's suggestion to split grading vertically rather than horizontally - questions rather than students - is good and practical irrespective of anything else.

That being said, I suggest you consider the following:

1. Give up the fantasy, or conceit that grading is consistent. It isn't, due to a zillion factors: Exams not of consistent difficulty; student population distribution of capabilities not in synch with grading curve shaping; sense of severity of errors differs for the same person for the first encounters and for latter encounters; moods change; unconscious biases (for example: against people with sloppy handwriting); and so on.
2. Try to schedule a grading policy discussion of all graders and the exam authors, before the next exam is given (next semester?). In that discussion, bring up some specific scenarios requiring finer judgement rather than simply "how many points for which question".
3. Consider arguing for rougher-granularity grading in the course. I'm a fan of pass/fail or pass/fail/excellent and dislike number scales, especially 0..100 or fractional grades.

These suggestions are mostly orthogonal, I think.

First, Ben's suggestion to split grading vertically rather than horizontally - questions rather than students - is good and practical irrespective of anything else.

That being said, I suggest you consider the following:

1. Give up the fantasy, or perhaps better put: the conceit, that grading is consistent. It isn't, due to a zillion factors: Exams not of consistent difficulty; student population distribution of capabilities not in sync with grading curve shaping; sense of severity of errors differs for the same person for the first encounters and for latter encounters; moods change; unconscious biases (for example: against people with sloppy handwriting); and so on.
2. Try to schedule a grading policy discussion of all graders and the exam authors, before the next exam is given (next semester?). In that discussion, bring up some specific scenarios requiring finer judgement rather than simply "how many points for which question".
3. Consider arguing for rougher-granularity grading in the course. I'm a fan of Pass/Fail (or Pass/Fail/Excellent), and dislike number scales, especially 0..100 or fractional grades. I can be confident about looking at someone and saying "Yeah, you understand what we've taught here, you pass." or "No, you don't get it - you fail." And if someone is borderline, then it's a matter of policy whether to pass or fail them (I would tend towards fail personally). But I really cannot justify why someone is a 63 while another person is a 64.7. I feel I'm only helping some arbitrary industrial mass-manipulation mechanism by assigning these kinds of numbers to people.

PS - These three suggestions are mostly orthogonal.